2686
Chem. Mater. 1996, 8, 2686-2692
Dynamic Quenching of Porous Silicon Excited States Minh C. Ko and Gerald J. Meyer* Department of Chemistry, Johns Hopkins University, Baltimore, Maryland 21218 Received April 25, 1996. Revised Manuscript Received July 19, 1996X
Porous silicon samples have been prepared from p-type single-crystal silicon 〈100〉 by a galvanostatic and an open-circuit etch in 50% HF. The materials display bright red-orange room-temperature photoluminescence (PL) in air and toluene solution. Infrared measurements show that the porous silicon surface is partially oxidized. Exposure to anthracene (An) or 10-methylphenothiazine (MPTZ) results in dynamic quenching of the material’s excited state(s). Nanosecond time-resolved PL decays are complex and wavelength dependent, with average lifetimes in neat toluene of 0.3-16 µs. Quenching by An and MPTZ is more efficient and rapid at short observation wavelengths. The steady-state and timeresolved quenching data are well fit to the Stern-Volmer model. The PL decays are well described by a skewed distribution of recombination rates.
Canham recently reported the preparation of silicon materials that display efficient, visible, room-temperature photoluminescence (PL).1 Silicon has long been the backbone of the electronics industry, but its indirect bandgap and weak PL properties have precluded applications in many optical devices. Canham’s discovery may one day change this, and an explosion of studies have been initiated to develop silicon-based electroluminescent devices from this material commonly referred to as porous silicon, po-Si.2 More fundamentally, researchers have probed the origin of the PL and its interesting physical and optical properties.2 An important goal is to develop a more complete understanding of the radiative and nonradiative decay pathways of poSi’s excited states. Toward this goal a number of researchers have used well-defined molecular species, such as organic compounds,3,8 amines,4 metal salts,5 and hydroxide ions,6 to probe the optical properties of poSi. The PL intensity from a semiconductor can be influenced by modulations in the surface electric field, the surface recombination velocity, or both. Even with welldefined single-crystal materials, it is experimentally difficult to quantify the underlying optoelectronic processes which drive PL intensity changes.8 The situation is even more complex for po-Si where the distribution of emissive states remains uncertain.2 The mechanism for quenching po-Si by molecular species cited above is therefore largely unknown. We recently reported that X Abstract published in Advance ACS Abstracts, September 1, 1996. (1) Canham, L. T. Appl. Phys. Lett. 1990, 57, 1046. (2) (a) Benshalet, D. C., et al., Eds. Optical Properties of Low Dimensional Silicon Structures; Kluwer Academic Publishers: Holland, 1993. (b) Brus, L. J. Phys. Chem. 1994, 98, 3575. (3) (a) Lauerhaas, J. M.; Credo, G. M.; Heinrich, J. L.; Sailor, M. J. J. Am. Chem. Soc. 1992, 114, 1911. (b) Lauerhaas, J. M.; Sailor, M. J. Science 1993, 261, 1567. (c) Fisher, D. L.; Harper, J.; Sailor, M. J. J. Am. Chem. Soc. 1995, 117, 7846. (4) Coffer, J. L.; Lilley, S. C.; Martin, R. A.; Files-Sesler, L. A. J. Appl. Phys. 1993, 74, 2094. (5) Andsager, D.; Hilliard, J.; Hetrick, J. M.; AbuHassan, L. H.; Plisch, M.; Nayfeh, M. H. J. Appl. Phys. 1993, 74, 4783. (6) Chun, J.; Bocarsly, A. B.; Cottrell, T. R.; Benziger, J. B.; Yee, J. C. J. Am. Chem. Soc. 1993, 115, 3024. (7) Ellis, A. B. In Chemistry and Structure at Interfaces: New Laser and Optical Techniques; Hall, R. B., Ellis, A. B., Eds.; VCH: Deerfield Beach, FL, 1986; Chapter 6. (8) Ko, M. C.; Meyer, G. J. Chem. Mater. 1995, 7, 12.
S0897-4756(96)00247-5 CCC: $12.00
exposure of 10-methylphenothiazine (MPTZ) or anthracene (An) to porous silicon in neat toluene quenches both the PL intensity and the average excited-state lifetime indicative of dynamic quenching.8 Further, the initial PL properties could be restored with toluene solvent. To our knowledge this represents the first demonstration of a reversible dynamic quenching mechanism for porous silicon excited states. The observation is important as it suggests that the surface recombination velocity at this technologically important interface can be reversibly tuned on a molecular level. Here we report more detailed studies of this dynamic quenching process. Experimental Section Materials. Toluene (Fisher Scientific), 50% HF (Fisher Scientific), and zone refined anthracene (Aldrich, 99+%) were used as received. 10-Methylphenothiazine (Pfaltz & Bauer) was recrystallized three times from toluene. Water was deionized with a Barnstead Nanopure system. Single crystal silicon wafers (〈100〉 0.5 orientation, p-type, boron doped, F ) 1 Ω cm) were obtained from Virginia Semiconductors, Inc. Sample Preparation. The Si wafers were cut into ∼3 cm2 sections using a diamond scribe, placed in an HF bath for ∼5 min and rinsed with deionized water (18 MΩ, Barnstead Nanopure System). The wafer was then etched galvanostatically at a current density of 13.4 mA/cm2 for 40 min in a 50 wt % HF solution in a custom-built Teflon two-chambered cell,9 washed with deionized water, and allowed to air dry for several minutes. The sample was then placed in the HF bath at open circuit for an additional 10-20 min, taken out, rinsed with deionized water, and dried. This procedure was repeated until intense red-orange PL was observed with UV light excitation. The samples were then stored in a desiccator until use. Prior to optical experiments, the samples were etched at open circuit in a 50% HF bath for 5-10 min. The samples were then dried and glued to a glass rod with Ducco cement, fitted in a thermometer adapter, and clamped into a 14/20 ground glass joint inside a custom-built glass cell. The cell has a Teflon stopcock at the bottom and a glass inlet at the top, which allows solution to be changed without altering the po-Si geometry. Optical Measurements. Steady-State PL. Corrected PL spectra were obtained with a Spex Fluorolog, Model 112A. The (9) (a) Friedersdorf, L. E.; Searson, P. C.; Prokes, S. M.; Glembocki, O. J.; Maccaulay, J. M. Appl. Phys. Lett. 1992, 60, 2285. (b) Searson, P. C.; Macaulay, J. M.; Ross, F. M. J. Appl. Phys. 1992, 72, 253.
© 1996 American Chemical Society
Dynamic Quenching of Porous Silicon Excited States
Chem. Mater., Vol. 8, No. 11, 1996 2687
Figure 2. Absorbance spectrum of anthracene (s) and 10methylphenothiazine (- - -) in toluene. The instrument resolution is (2 nm. Figure 1. Schematic diagram of the apparatus employed to obtain kinetic and gated PL spectra of po-Si materials. A detailed description is given in the Experimental Section.
UV-Vis. UV-vis absorbance measurements were made on a Hewlett-Packard 8451A diode array, with (2 nm and (0.005 au resolution.
PL was obtained in a front-face mode with a Hamamatsu R668 PMT optically coupled to a Spex dual-grating monochromator. The PL spectra were corrected with a calibration curve generated with 2.0 mm slits by using a NBS calibrated 250 W lamp (Optronics Laboratories, Inc. Model 220M), controlled by a precision current source at 6.50 W (Optronics Laboratories, Inc. Model 65). For excitation spectra, a beam splitter redirects a portion of the excitation light into a rhodamine B quantum counter to monitor the wavelength-dependent irradiance of the Xe lamp. The excitation spectra is recorded as the ratio between the PL intensity and the excitation irradiance. Time Resolved PL. Time-resolved PL studies were performed with the apparatus shown in Figure 1. A Laser Photonics LN 100/107 nitrogen-pumped dye laser with Coumarin 460 (exciton) was used for excitation. Typical excitation irradiance was 40-60 µJ/pulse monitored with a Molectron J3-09 joulemeter at a repetition rate of ∼1 Hz. The PL was collected with two f/2 lenses to a McPherson Model 272 f/2 scanning monochromator. A long-pass 495 filter was placed between the sample and the monochromator to remove scattered light. A Hamamatsu R928 photomultiplier tube (PMT) mounted in an EMI-shielded housing was optically coupled to the monochromator. The PMT base was wired for fast response and biased at -900 V with a Thorn EMI Model 3000R power supply. The signal from the PMT was terminated into 50 Ω of a LeCroy 9450 350 MHz digital storage oscilloscope optically triggered with a photodiode (Motorola MRD 500) mounted in the dye laser. The instrument response function was measured to be 14 ns with colloidal SiO2 (LUDOX) as a light scatterer. The extrema and average of 100-200 decays were collected and subsequently transferred to a 486 microprocessor via a GPIB bus. For data fitting, the time base was adjusted such that the initial PL response occurs at time zero and data analysis was initiated at 30 ns. The data were analyzed by a Marquardt algorithm and/or a Nelder-Mead modified simplex routine with code written in ASYST (Keithly). Alternatively, a fiber optic link (FOL) brought the excitation to the sample positioned in front of an Instruments SA HR320 spectrograph coupled to a Princeton Instruments intensified charge-coupled device (ICCD). The PL was collected with a 50 mm camera lens. The ICCD was controlled with a Princeton Instruments Controller and a PG-200 pulse generator. Data were sent from the Controller to a 486 or 586 microprocessor via a direct memory access (DMA) cable. The ICCD was calibrated with a Hg(Ar) lamp (Oriel). Infrared. IR spectra were taken on a Perkin-Elmer 1600 Series FTIR ((4 cm-1) instrument in a transmission mode. The sample compartment was purged with dry N2 gas during all experiments.
Results The absorbance spectra of anthracene (An) and 10methylphenothiazine (MPTZ) in toluene is given in Figure 2. At wavelengths longer than 400 nm there is negligible absorbance, which indicates that the organic compounds do not absorb light at the excitation or observation wavelengths. Porous silicon samples were prepared by a galvanostatic etch in 50% HF for 40 min at a current density of 13.4 mA/cm2. A significant change from our previous studies is that an open-circuit etch in 50 wt % HF was performed immediately before optical measurements. The open circuit etch is known to increase the PL intensity and porosity of the po-Si.9 FTIR spectra of poSi before and after the steady-state quenching studies with An and MPTZ described below are typical of partially oxidized porous silicon and the absorbances can be assigned based on previously reported studies.10 Spectral features near 1100 cm-1 are assigned to SiO-Si.10a There are three very distinct peaks centered at 2137, 2110, and 2088 cm-1 and a peak at 907 cm-1 which have been ascribed to SiHx vibrations.10b IR measurements before and after PL quenching experiments showed a slight increase in intensity of the signal near 1100 cm-1 with no significant changes in the (SiHx) modes near 2100 cm-1. Upon illumination of po-Si in toluene with ∼1-2 mW/ cm2 of 460 nm light, a gradual decrease in PL intensity is observed over the first 20-30 min. After this time a stable PL intensity is observed with only minor changes in intensity after hours of illumination. Steady-state measurements were generally performed after this illumination/conditioning period. The corrected PL and excitation spectrum of a representative sample in neat toluene is shown in Figure 3. The corrected PL maximum is typically 710 ( 30 nm and the full width at half maximum is 2500 ( 200 cm-1 with 460 nm excitation. When the excitation wavelength is decreased, the PL maximum shifts toward higher energy. For example, the PL maximum shifts from 695 to 760 nm when the (10) (a) Silverstein, R. M.; Bassler, G. C.; Morrill, T. C. in Spectrometric Identification of Organic Compounds; John Wiley & Sons: New York, 1981; pp 175. (b) Brodsky, M. H.; Cardona, M.; Cuomo, J. J. Phys. Rev. B 1977, 16, 3556.
2688 Chem. Mater., Vol. 8, No. 11, 1996
Figure 3. Corrected room-temperature PL (inset) and excitation spectra of po-Si in neat toluene. The excitation spectra was recorded at λmon ) 735 nm and the PL spectra was recorded at λex ) 460 nm. The incident irradiance at 460 nm was 1.79 mW/cm2 and all wavelengths are (3 nm.
Figure 4. (a) PL spectra of a po-Si sample as a function of anthracene (An) concentration in toluene. The millimolar concentrations are (A) ) 0.00; (B) ) 0.60, (C) ) 1.05, (D) ) 1.40, (E) 1.68, (F) ) 1.91; and (G) ) 2.10. (b) PL spectra of a po-Si sample as a function of 10-methylphenothiazine (MPTZ) concentration in toluene. The millimolar concentrations are (A) ) 0.00, (B) 0.06, (C) 0.13, (D) 0.20, (E) 0.26, (F) 0.54, (G) 0.78, and (H) 0.99. The excitation wavelength was 460 ( 3 nm.
excitation wavelength is moved from 420 to 500 nm. The incident irradiance at 420, 460, and 500 nm was 1.68, 1.79, and 1.47 mW/cm2, respectively. PL quenching as a function of quencher concentration with 460 nm excitation is shown in Figure 4. Addition of An or MPTZ results in a reversible decrease in the PL intensity and red-shift in the PL spectra. Excitation
Ko and Meyer
Figure 5. PL intensity of a po-Si sample with alternate exposures to neat toluene (initial intensity) and to 4 mM An or 4 mM MPTZ recorded in a time-base mode. The extent of quenching was ∼15% for An and ∼66% for MPTZ. The sample was excited with 460 ( 3 nm and monitored at 730 ( 3 nm.
at shorter wavelengths leads to a more dramatic redshift with added quencher than does excitation with longer wavelengths of light. In our previous studies, with 475 nm light excitation the spectral shift with added quencher could not be resolved, for example.8 The data shown in Figure 4 were collected with 460 nm excitation, and the spectral shift is clearly seen. With 400 nm excitation, addition of quencher results in a more pronounced red shift. All the steady-state and time-resolved data reported here were obtained with 460 nm excitation so that internal comparisons could be made and to avoid the possibility of direct excitation of the quenchers. To explore the reproducibility and reversibility of the quenching processes, experiments were performed where the PL intensity from an individual po-Si sample was monitored in neat toluene and toluene-quencher solutions in a time-base mode. Shown in Figure 5 are typical results from a po-Si sample monitored at 730 nm. The sample is initially in neat toluene and is then exposed to 4 mM An which results in a 15% decrease in PL intensity. The solution is then changed to 4 mM An a second time to demonstrate the reproducibility of the quenching. The sample is then exposed to 4 mM MPTZ and a more dramatic ∼66% decrease in PL intensity is reproducibly observed. Neat toluene restores the PL intensity to within 95% of the initial value. Studies with a large number of samples have shown that continued cycling between neat toluene and toluene-quencher solutions results in an increased degree of reversibility. Time-resolved PL decays from porous silicon in neat toluene are complex and dependent on the monitoring wavelength. With 50 ( 10 µJ/pulse excitation at 460 nm the decays were nonexponential at all monitoring wavelengths and quencher concentrations studied. Normalized and unnormalized PL decays were analyzed by the Kohlrausch-Williams-Watts function (KWW)11 which has been used extensively for po-Si materials, (eq 1).12 Here R is the initial PL intensity, β is related to the width of an underlying Levy distribution of rates,
PLI(t) ) R exp(-t/τ)β
0
